Merge pull request #163 from inariksit/persian-fixes

Persian fixes

src/persian/CatPes.gf

@@ -53,7 +53,7 @@ concrete CatPes of Cat = CommonX - [Adv] ** open ResPes, Prelude in {
 
 ---- Noun
 
-    CN = ResPes.Noun ;
+    CN = ResPes.CN ;
 
     NP = ResPes.NP ;
     Pron = {s : Str ; ps : Str ; a : Agr};

src/persian/ConjunctionPes.gf

@@ -12,7 +12,7 @@ concrete ConjunctionPes of Conjunction =
     ConjAdv = conjunctDistrSS ;
 --    ConjAdv conj advs = conjunctDistrTable Gender conj advs ;
 
-    ConjNP conj ss = conjunctDistrTable NPCase conj ss ** {
+    ConjNP conj ss = ss ** conjunctDistrTable NPCase conj ss ** {
       a = conjAgr (agrP3 conj.n) ss.a ;
       animacy = ss.animacy ;
       } ;
@@ -28,9 +28,9 @@ concrete ConjunctionPes of Conjunction =
 --    BaseAdv x y = twoTable Gender x y  ;
     ConsAdv = consrSS comma ;
 --    ConsAdv xs x = consrTable Gender comma xs x ;
-    BaseNP x y = twoTable NPCase x y ** {a = conjAgr x.a y.a ; animacy = y.animacy } ; -- check animacy
+    BaseNP x y = y ** twoTable NPCase x y ** {a = conjAgr x.a y.a ; animacy = y.animacy } ; -- check animacy
     BaseRS x y = twoTable Agr x y ** {c = x.c};
-    ConsNP xs x = consrTable NPCase comma xs x ** {a = conjAgr xs.a x.a ; animacy = xs.animacy } ; --  InaandB xs.animacy x.animacy} ;
+    ConsNP xs x = xs ** consrTable NPCase comma xs x ** {a = conjAgr xs.a x.a ; animacy = xs.animacy } ; --  InaandB xs.animacy x.animacy} ;
     ConsRS xs x = consrTable Agr comma xs x ** { c = xs.c};
 --    BaseAP x y = twoTable3 Number Gender Case x y ; -- ** {isPre = andB x.isPre y.isPre} ;
     BaseAP x y = twoTable Ezafa x y ** {adv = x.adv};
@@ -39,7 +39,7 @@ concrete ConjunctionPes of Conjunction =
   lincat
     [S] = {s1,s2 : Str} ;
     [Adv] = {s1,s2 : Str} ;
-    [NP] = {s1,s2 : NPCase => Str ; a : Agr ; animacy : Animacy } ;
+    [NP] = {s1,s2 : NPCase => Str ; a : Agr ; animacy : Animacy ; hasAdj : Bool} ;
     [AP] = {s1,s2 :  Ezafa => Str ; adv : Str} ;
     [RS] = {s1,s2 : Agr => Str };
 

src/persian/MorphoPes.gf

@@ -55,23 +55,24 @@ oper
 oper
 
   mkEzafa : Str -> Str ;
-  mkEzafa str = case str of {
+  mkEzafa str = let kasre = "ِ" in
-      st + "اه" => str ;
+    case str of {
-      st + "وه" => str ;
+      st + "اه" => str + kasre ;
+      st + "وه" => str + kasre ;
       st + "ه"  => st + "ۀ" ; -- str ++ "ی" ;
-      st + "او" => str ;
+      st + "او" => str + kasre ;
-      st + "وو" => str ;
+      st + "وو" => str + kasre ;
       st + "و"  => str + "ی" ;
       st + "ا"  => str + "ی" ;
-      _ => str
+      _ => str + kasre
     };
+
   mkEnclic : Str -> Str ;
   mkEnclic str = case str of {
-      st + "ا" => str ++ "یی" ;
+      st + ("ا"|"و") => zwnj str "یی" ; -- ی after a long vowel to help pronunciation
-      st + "و" => str ++ "یی" ;
+      st + "اه"      =>      str + "ی" ; -- here ه is a consonant, so single ی
-      st + "ی" => str ++ "یی" ; -- TODO
+      st + ("ی"|"ه") => zwnj str "ای" ; -- after ی or ه as a vowel, add an alif to help pronunciation
-      st + "ه" => str ++ "یی" ;
+      _               =>      str + "ی" -- any other case: just a single ی
-      _        => str + "ی"
     } ;
 
   Noun = {s : Ezafa => Number => Str ; animacy : Animacy ; definitness : Bool } ;

src/persian/NounPes.gf

@@ -3,19 +3,18 @@ concrete NounPes of Noun = CatPes ** open ResPes, Prelude in {
   flags optimize=all_subs ;
 
   lin
-    DetCN det cn = { s = \\_ =>
+    DetCN det cn = cn ** {s = \\npc =>
-      case <det.isNum,det.fromPron> of {
+      case <det.isNum,det.fromPron,npc> of {
-        <False,True>  => cn.s ! aEzafa ! det.n ++ det.s ;
+        <False,False,NPC ez> => det.s ++ cn.s ! ez ! det.n ; -- det is not from Pron, retain NPForm.
-        <False,False> => det.s ++ cn.s ! bEzafa ! det.n ;
+        <False,True>         => cn.s ! aEzafa ! det.n ++ det.s ; -- det is from Pron, cn is in aEzafa.
-        <True,True>   => cn.s ! aEzafa ! Sg ++ det.s ; -- noun modified by a number is invariably singular
+        <True,False,NPC ez>  => det.s ++ cn.s ! ez ! Sg ; -- noun modified by a number is invariably singular
-        <True,False>  => det.s ++ cn.s ! bEzafa ! Sg
+        <True,True>          => cn.s ! aEzafa ! Sg ++ det.s
       } ;
       a = agrP3 det.n ;
-      animacy = cn.animacy
       } ;
 
-    UsePN pn = {s = \\_ => pn.s ; a = agrP3 Sg ; animacy = pn.animacy } ;
+    UsePN pn = pn ** {s = \\_ => pn.s ; a = agrP3 Sg ; hasAdj = False} ;
-    UsePron p = {s = \\_ => p.s ; a = p.a ; animacy = Animate} ;
+    UsePron p = p ** {s = \\_ => p.s ; animacy = Animate ; hasAdj = False} ;
 
     PredetNP pred np = np ** {
       s = \\ez => pred.s ++ np.s ! ez
@@ -50,6 +49,7 @@ concrete NounPes of Noun = CatPes ** open ResPes, Prelude in {
     DetNP det = {
       s = \\_ => det.s  ; ---- case
       a = agrP3 det.n ;
+      hasAdj = False ;
       animacy = Inanimate
       } ;
 
@@ -73,46 +73,41 @@ concrete NounPes of Noun = CatPes ** open ResPes, Prelude in {
     DefArt = {s = \\_ => [] ; a = defaultAgr ; fromPron = False} ;
     IndefArt = {s = table { Sg => IndefArticle ; Pl => []} ; a =defaultAgr ; fromPron = False} ;
 
-    MassNP cn = {s =\\c => case c of {
+    MassNP cn = cn ** {s =\\c => case c of {
       NPC bEzafa => cn.s ! bEzafa ! Sg ;
       NPC aEzafa => cn.s ! aEzafa ! Sg ;
       NPC enClic => cn.s ! enClic ! Sg
       };
       a = agrP3 Sg ;
-      animacy = cn.animacy
       } ;
 
-    UseN n = n ;
+    UseN n = n ** {hasAdj=False};
-    UseN2 n = n ;
+    UseN2 n = n ** {hasAdj=False};
 
-    Use2N3 f = {
+    Use2N3 f = f ** {
-      s = f.s;
       c = f.c2;
-      animacy = f.animacy;
       definitness = True
       } ;
 
-   Use3N3 f = {
+   Use3N3 f = f ** {
-      s = f.s ;
       c = f.c3;
-      animacy = f.animacy;
       definitness = True
       } ;
 
-    ComplN2 f x = {
+    ComplN2 f x = f ** {
       s = \\ez,n => f.s ! ez ! n  ++ f.c ++ x.s ! NPC ez ;
-      animacy = f.animacy;
+      definitness = True ;
-      definitness = True
+      hasAdj = False ;
      };
-    ComplN3 f x = {
+    ComplN3 f x = f ** {
       s = \\ez,n => f.s ! ez ! n ++ f.c2 ++ x.s ! NPC ez ;
       c = f.c3;
-      animacy = f.animacy;
       definitness = True;
       } ;
 
     AdjCN ap cn = cn ** {
-      s = \\ez,n =>  cn.s ! aEzafa ! n ++ ap.s ! ez -- check the form of adjective and also cn.s!ez!n changed from cn.s!aEzafa!n to have correct enclicitic form other wise it creats wrong enclictic form of old man
+      s = \\ez,n => cn.s ! aEzafa ! n ++ ap.s ! ez ; -- check the form of adjective and also cn.s!ez!n changed from cn.s!aEzafa!n to have correct enclicitic form other wise it creats wrong enclictic form of old man
+      hasAdj = True
      } ;
 
     RelCN cn rs = cn ** {

src/persian/ResPes.gf

@@ -25,7 +25,8 @@ resource ResPes = MorphoPes ** open Prelude,Predef in {
   oper
 
     Compl : Type = {s : Str ; ra : Str ; c : VType} ;
-    NP : Type = {s : NPCase => Str ; a : Agr ; animacy : Animacy } ;
+    CN : Type = Noun ** {hasAdj : Bool} ; -- for getting the right form when NP/CN is a predicate
+    NP : Type = {s : NPCase => Str ; a : Agr ; animacy : Animacy ; hasAdj : Bool} ;
     VPHSlash : Type = VPH ** {c2 : Compl} ;
 
   oper
@@ -34,6 +35,11 @@ resource ResPes = MorphoPes ** open Prelude,Predef in {
       Neg => CNeg b
     } ;
 
+    cpol2pol : CPolarity -> Polarity = \cp -> case cp of {
+      CPos => Pos ;
+      _    => Neg
+    } ;
+
  -----------------------
  --- Verb Phrase
  -----------------------
@@ -144,7 +150,7 @@ oper
 
   insertVV : (Agr => Str) -> VPH -> VPH = \obj1,vp -> vp ** {
     wish = True ;
-    vComp = \\a => vp.comp ! a ++ obj1 ! a ;
+    vComp = \\a => vp.comp ! a ++ conjThat ++ obj1 ! a ;
   } ;
 
   insertObj2 : (Str) -> VPH -> VPH = \obj1,vp -> vp ** {

src/persian/SentencePes.gf

@@ -10,12 +10,12 @@ concrete SentencePes of Sentence = CatPes ** open Prelude, ResPes,Predef in {
     PredSCVP sc vp = mkSClause ("این" ++ sc.s) (defaultAgr) vp ;
 
     ImpVP vp = {
-      s = \\pol,n =>
+      s = \\cpol,n =>
         let agr = Ag (numImp n) P2 ;
-        in case pol of {
+            pol = cpol2pol cpol ;
-             CPos   => vp.ad ++ vp.comp ! agr ++ vp.obj.s ++ vp.vComp ! agr ++ vp.s ! VPImp Pos (numImp n) ++ vp.embComp;
+         in case vp.wish of {
-             CNeg _ => vp.ad ++ vp.comp ! agr ++ vp.obj.s ++ vp.vComp ! agr ++ vp.s ! VPImp Neg (numImp n) ++ vp.embComp
+              True  => vp.s ! VPImp pol (numImp n) ++ vp.ad ++ vp.comp ! agr ++ vp.obj.s ++ vp.vComp ! agr ++ vp.embComp;
-	  } ;
+              False => vp.ad ++ vp.comp ! agr ++ vp.obj.s ++ vp.vComp ! agr ++ vp.s ! VPImp pol (numImp n) ++ vp.embComp }
     } ;
 
 

src/persian/SymbolPes.gf

@@ -10,20 +10,17 @@ concrete SymbolPes of Symbol = CatPes ** open Prelude, ResPes in {
   IntPN i  = {s = i.s ; animacy = Inanimate} ;
   FloatPN i = {s = i.s ; animacy = Inanimate} ;
   NumPN i = {s = i.s ; animacy = Inanimate} ;
-  CNIntNP cn i = {
+  CNIntNP cn i = cn ** {
     s = \\ez => cn.s ! aEzafa ! Sg  ++ i.s ;
-    a = agrP3 Sg ;
+    a = agrP3 Sg
-    animacy = cn.animacy
     } ;
-  CNSymbNP det cn xs = {
+  CNSymbNP det cn xs = cn ** {
     s = \\ez => det.s ++ cn.s ! aEzafa ! det.n ++ xs.s ;
-    a = agrP3 det.n ;
+    a = agrP3 det.n
-    animacy = cn.animacy
     } ;
-  CNNumNP cn i = {
+  CNNumNP cn i = cn ** {
     s = \\ez => cn.s ! aEzafa ! Sg ++ i.s ;
-    a = agrP3 Sg ;
+    a = agrP3 Sg
-    animacy = cn.animacy
     } ;
 
   SymbS sy = sy ;

src/persian/VerbPes.gf

@@ -1,4 +1,4 @@
-concrete VerbPes of Verb = CatPes ** open ResPes in {
+concrete VerbPes of Verb = CatPes ** open ResPes,Prelude in {
 
   flags coding = utf8;
   flags optimize=all_subs ;
@@ -42,9 +42,19 @@ concrete VerbPes of Verb = CatPes ** open ResPes in {
     PassV2 v = predV v ; -- need to be fixed
     CompAP ap ={s = \\_ => ap.s ! bEzafa} ; -- check form of adjective
     CompAdv adv = {s = \\_ => adv.s } ;
-    CompCN cn = {s = \\a => cn.s ! bEzafa !  giveNumber a } ;
+
-    -- IL 2019-01-31 changed bEzafa to enClic according to
+    -- see https://sites.la.utexas.edu/persian_online_resources/nouns/noun-in-a-predicative-position/
-    -- https://sites.la.utexas.edu/persian_online_resources/nouns/noun-in-a-predicative-position/
+    CompCN cn = {
-    CompNP np = {s = \\a => np.s ! NPC enClic} ;
+      s = \\a => cn.s ! case cn.hasAdj of {
+                          False => bEzafa ;
+                          True  => enClic }
+                      ! giveNumber a
+      } ;
+
+    CompNP np = {
+      s = \\a => np.s ! case np.hasAdj of {
+                          False => NPC bEzafa ;
+                          True  => NPC enClic }
+      } ;
 
 }